From 5df9d958abe7dc96e849e201fd4f121657eba932 Mon Sep 17 00:00:00 2001
From: Andrew Noyes <andrew@weaselab.dev>
Date: Mon, 23 Jun 2025 14:00:29 -0400
Subject: [PATCH] Unroll whitespace-skipping loop

---
 src/parser3.h | 77 ++++++++++++++++++++++++---------------------------
 1 file changed, 36 insertions(+), 41 deletions(-)

diff --git a/src/parser3.h b/src/parser3.h
index c8636cd..412b4ce 100644
--- a/src/parser3.h
+++ b/src/parser3.h
@@ -750,17 +750,33 @@ struct Parser3 {
   Utf8Dfa strDfa;
 };
 
+inline PRESERVE_NONE WeaselJsonStatus skipWhitespace(char *&buf, char *bufEnd) {
+  constexpr int kStride = 4;
+  for (;;) {
+    if (bufEnd - buf < kStride) [[unlikely]] {
+      while (buf != bufEnd && tables.whitespace[uint8_t(*buf)]) {
+        ++buf;
+      }
+      return buf == bufEnd ? WeaselJson_AGAIN : WeaselJson_OK;
+    }
+    for (int i = 0; i < kStride; ++i) {
+      if (tables.whitespace[uint8_t(*buf)]) {
+        ++buf;
+      } else {
+        return WeaselJson_OK;
+      }
+    }
+  }
+}
+
 inline PRESERVE_NONE WeaselJsonStatus n_whitespace(Parser3 *self, char *buf,
                                                    char *bufEnd) {
-  if (bufEnd - buf == 0) {
+  if (buf == bufEnd) {
     self->pop();
     MUSTTAIL return Parser3::keepGoing(self, buf, bufEnd);
   }
-  while (tables.whitespace[uint8_t(*buf)]) {
-    ++buf;
-    if (buf == bufEnd) {
-      return WeaselJson_AGAIN;
-    }
+  if (auto s = skipWhitespace(buf, bufEnd)) {
+    return s;
   }
   self->pop();
   MUSTTAIL return Parser3::keepGoing(self, buf, bufEnd);
@@ -856,11 +872,8 @@ inline PRESERVE_NONE WeaselJsonStatus scan_string(Parser3 *self, char *buf,
 inline PRESERVE_NONE WeaselJsonStatus n_value(Parser3 *self, char *buf,
                                               char *bufEnd) {
   assert(bufEnd - buf != 0);
-  while (tables.whitespace[uint8_t(*buf)]) {
-    ++buf;
-    if (buf == bufEnd) {
-      return WeaselJson_AGAIN;
-    }
+  if (auto s = skipWhitespace(buf, bufEnd)) {
+    return s;
   }
   switch (*buf) {
   case '{':
@@ -963,11 +976,8 @@ inline PRESERVE_NONE WeaselJsonStatus n_value(Parser3 *self, char *buf,
 inline PRESERVE_NONE WeaselJsonStatus n_object2(Parser3 *self, char *buf,
                                                 char *bufEnd) {
   assert(bufEnd - buf != 0);
-  while (tables.whitespace[uint8_t(*buf)]) {
-    ++buf;
-    if (buf == bufEnd) {
-      return WeaselJson_AGAIN;
-    }
+  if (auto s = skipWhitespace(buf, bufEnd)) {
+    return s;
   }
   switch (*buf) {
   case '}':
@@ -992,11 +1002,8 @@ inline PRESERVE_NONE WeaselJsonStatus n_object2(Parser3 *self, char *buf,
 inline PRESERVE_NONE WeaselJsonStatus n_object3(Parser3 *self, char *buf,
                                                 char *bufEnd) {
   assert(bufEnd - buf != 0);
-  while (tables.whitespace[uint8_t(*buf)]) {
-    ++buf;
-    if (buf == bufEnd) {
-      return WeaselJson_AGAIN;
-    }
+  if (auto s = skipWhitespace(buf, bufEnd)) {
+    return s;
   }
   switch (*buf) {
   case '}':
@@ -1019,11 +1026,8 @@ inline PRESERVE_NONE WeaselJsonStatus n_object3(Parser3 *self, char *buf,
 inline PRESERVE_NONE WeaselJsonStatus n_array2(Parser3 *self, char *buf,
                                                char *bufEnd) {
   assert(bufEnd - buf != 0);
-  while (tables.whitespace[uint8_t(*buf)]) {
-    ++buf;
-    if (buf == bufEnd) {
-      return WeaselJson_AGAIN;
-    }
+  if (auto s = skipWhitespace(buf, bufEnd)) {
+    return s;
   }
   switch (*buf) {
   case ']':
@@ -1043,11 +1047,8 @@ inline PRESERVE_NONE WeaselJsonStatus n_array2(Parser3 *self, char *buf,
 inline PRESERVE_NONE WeaselJsonStatus n_array3(Parser3 *self, char *buf,
                                                char *bufEnd) {
   assert(bufEnd - buf != 0);
-  while (tables.whitespace[uint8_t(*buf)]) {
-    ++buf;
-    if (buf == bufEnd) {
-      return WeaselJson_AGAIN;
-    }
+  if (auto s = skipWhitespace(buf, bufEnd)) {
+    return s;
   }
   switch (*buf) {
   case ']':
@@ -1070,11 +1071,8 @@ inline PRESERVE_NONE WeaselJsonStatus n_array3(Parser3 *self, char *buf,
 inline PRESERVE_NONE WeaselJsonStatus n_string(Parser3 *self, char *buf,
                                                char *bufEnd) {
   assert(bufEnd - buf != 0);
-  while (tables.whitespace[uint8_t(*buf)]) {
-    ++buf;
-    if (buf == bufEnd) {
-      return WeaselJson_AGAIN;
-    }
+  if (auto s = skipWhitespace(buf, bufEnd)) {
+    return s;
   }
   if (*buf != '"') [[unlikely]] {
     return WeaselJson_REJECT;
@@ -1318,11 +1316,8 @@ inline PRESERVE_NONE WeaselJsonStatus singleChar(Parser3 *self, char *buf,
                                                  char *bufEnd) {
   if constexpr (kSkipWhitespace) {
     assert(bufEnd - buf != 0);
-    while (tables.whitespace[uint8_t(*buf)]) {
-      ++buf;
-      if (buf == bufEnd) {
-        return WeaselJson_AGAIN;
-      }
+    if (auto s = skipWhitespace(buf, bufEnd)) {
+      return s;
     }
   }
   if (*buf == kChar) {